Subject: port-macppc/33605: kernel trap during page
To: None <port-macppc-maintainer@netbsd.org, gnats-admin@netbsd.org,>
From: None <scott@illogics.org>
List: netbsd-bugs
Date: 05/30/2006 20:15:00
	Note: There was a bad value `' for the field `Class'.
	It was set to the default value of `sw-bug'.

	Note: There was a bad value `' for the field `Priority'.
	It was set to the default value of `medium'.

	Note: There was a bad value `' for the field `Severity'.
	It was set to the default value of `serious'.

>Number:         33605
>Category:       port-macppc
>Synopsis:       kernel trap during page
>Confidential:   no
>Severity:       serious
>Priority:       medium
>Responsible:    port-macppc-maintainer
>State:          open
>Class:          sw-bug
>Submitter-Id:   net
>Arrival-Date:   Tue May 30 20:15:00 +0000 2006
>Originator:     Scott Walters
>Release:        NetBSD 3.0
>Organization:
	
>Environment:
System: 
NetBSD blacklight.slowass.net 3.0 NetBSD 3.0 (GENERIC) #0: Mon Dec 19 01:25:38 UTC 2005  builds@b2.netbsd.org:/home/builds/ab/netbsd-3-0-RELEASE/macppc/200512182024Z-obj/home/builds/ab/netbsd-3-0-RELEASE/src/sys/arch/macppc/compile/GENERIC macppc
Architecture: powerpc
Machine: macppc

>Description:

Crash during moderate use (one user reported cp'ing files) with the following messages on console:

mc0: underflow
mc0: underflow
trap: kernel read DSI trap @ 0x7ff0528c by 0x238c18 (DSISR 0x40000000, err=14)
panic: trap
Stopped in pid 6.1 (pagedaemon) at    netbsd:cpu_Debugger+0x10:   lwz r0, r1, 0x14

This came... let me count... one, two... two days after upgrading from 1.6.2.  God, oh, god,
oh god... please don't EOL 1.6.x.  I don't care about the fiddly little bug fixes, only
the major crashers introduced.  All I want is stability.  I'll go buy hardware to run
a stable operating system.  I don't care what you support.  Damn kids these days don't
know what "stable" means.  Knock it off with the code churn.  A lot of us use NetBSD 
exactly because people aren't in there futzing with stuff all the time.  A big changelog
is *not* a feature.  And while I'm on it, what's the deal with PAM?  Do you have any
idea how many "Linux vulnerabilities" are actually PAM vulnerabilities?  Adopting PAM
at this point is like taking in your cocain adopted 15 year old son because he's 
done so many drugs, he couldn't *possibily* do any more and *must* have cleaned up his
act by now.  

If you guys want to dabble in OS internals, why not do up a nice microkernel so these
funky algorithms have a sandbox to run in?  

Seriously though... when I can't find a free operating system that can survive two days
of moderate use, clients start forcing me to do things like run SCO.  That makes me
extremely unhappy.  

	
>How-To-Repeat:
	
>Fix:
	

>Unformatted: